We're releasing the weights of Mistral Large 4

Mistral has opened access to a model with roughly one trillion parameters, according to a daily AI roundup. Releasing the weights lets teams inspect the model, fine-tune it and run it on their own hardware, subject to the license terms in Mistral's announcement. Open-weight releases at this scale give enterprises a self-hosted option that does not depend on a single hosted API provider. That matters for regulated sectors where sending prompts and documents to an outside service is a hard limit, and it gives European buyers a frontier-scale option from a domestic lab. For builders, the practical question is hardware. A model this size needs multi-node serving, so smaller teams may use it as a teacher for distillation or as a reference point rather than for production traffic. Benchmark it against smaller open models on your own tasks before committing infrastructure budget.